[Python] Pandas apply function behaving differently based on input size?

I have a function which works fine with a tiny Pandas Dataframe and returns the adjustments as expected, but when I apply it to a non-test Dataframe, which is only a small df (300 x 20), it gets completely messed up.

Using the example below as a reference, it would populate columns 3 - 7 with the same list of values for all rows in each respective column...

To add to the confusion, when I remove the code from within the function and run it as part of one file the expected results are achieved for the non-test dataframe...

Can anyone suggest what might be happening here and why the behaviour seems to vary? / what I am changing.

I provide the code here for recreation

def test_populate_columns_from_Key_Value_String():

starting_df = pd.DataFrame(
{
"Names": ["Sarah", "John"],
"Relations": [
"has parent: Maggie, is parent of: Tom, is parent of: Grace, is parent of: Bart, is related to: Grandpa Simpson, is hated by: Joseph",
"is friends with: Tracey, has parent: Greg",
],
"has parent": [[], []],
"is parent of": [[], []],
"is related to": [[], []],
"is hated by": [[], []],
"is friends with": [[], []],
}
)

desire_df = pd.DataFrame(
{
"Names": ["Sarah", "John"],
"Relations": [
"has parent: Maggie, is parent of: Tom, is parent of: Grace, is parent of: Bart, is related to: Grandpa Simpson, is hated by: Joseph",
"is friends with: Tracey, has parent: Greg",
],
"has parent": [["Maggie"], ["Greg"]],
"is parent of": [["Tom", "Grace", "Bart"], []],
"is related to": [["Grandpa Simpson"], []],
"is hated by": [["Joseph"], []],
"is friends with": [[], ["Tracey"]],
}
)

starting_df.apply(
populate_columns_from_Key_Value_String, axis=1, args=("Relations",)
)

print(starting_df)
assert desire_df.loc[0, "has parent"] == starting_df.loc[0, "has parent"]
assert starting_df.loc[0, "is hated by"] == ["Joseph"]
assert starting_df.loc[1, "is friends with"] == ["Tracey"]
assert starting_df.loc[0, "is parent of"] == desire_df.loc[0, "is parent of"]

def populate_columns_from_Key_Value_String(
row: pd.Series, column_to_extract_from="Column_name_with_Relational_String"
):
string_to_extract_values_from = row[column_to_extract_from]
if isinstance(string_to_extract_values_from, str):
for key_value_pair in string_to_extract_values_from.split(","):
if isinstance(row[key_value_pair.split(":")[0].strip()], list):
row[key_value_pair.split(":")[0].strip()].append(
key_value_pair.split(":")[1].strip()
)

Continue reading...